Steven Rosenbaum, author of the recently released and much-hyped book, “The Future of Truth,” found himself in an awkward predicament this week when he publicly admitted to incorporating numerous fabricated or misattributed quotes, concocted by artificial intelligence, into his work. The revelation came after The New York Times, in reviewing sections of the book, identified more than half a dozen such quotes and questioned Mr. Rosenbaum. On Monday night, he issued a statement acknowledging “a handful of improperly attributed or synthetic quotes” and committed to launching his own investigation. He maintained that these inclusions were accidental, stating he had “no intention of fabricating any viewpoints” and disclosed that he had utilized AI tools like ChatGPT and Claude during his research, writing, and editing processes. While taking full responsibility, he emphasized this did not excuse the errors and promised to work with editors to swiftly correct any affected passages for future editions. “The Future of Truth” was published by an imprint of BenBella Books and distributed by Simon and Schuster, both of whom have remained silent on the matter.
Mr. Rosenbaum is a familiar face in the media industry, serving as the executive director of the Sustainable Media Center, a nonprofit dedicated to empowering a new generation of media consumers and creators. The center has a history of bringing together influential figures in media and technology for various events. His book had already garnered significant attention, with an excerpt appearing in Wired magazine and glowing endorsements from prominent journalists such as Taylor Lorenz, Michael Wolff, and Nicholas Thompson. The foreword was even penned by Maria Ressa, a Nobel Peace Prize winner renowned for her critical examination of Rodrigo Duterte, the former president of the Philippines. This incident, however, underscores a growing concern among publishers about authors inadvertently—or intentionally—using AI-generated content, a fear that saw Hachette withdraw a forthcoming horror novel this year due to similar allegations.
While many of the quotes in Mr. Rosenbaum’s book are accurate, the problematic misattributed and invented quotes are sprinkled throughout. For instance, in a chapter discussing AI lies, a quote is attributed to prominent technology journalist Kara Swisher: “The most sophisticated A.I. language model is like a mirror. It reflects our own morality back at us, polished and articulate, but ultimately empty behind the surface. It’s not bound by Asimov’s laws or any ethical framework — it’s bound by the patterns in its training data and the objectives set by its creators.” Upon being asked about this, Ms. Swisher emphatically stated in a text message that she “never said that,” and humorously added that it seemed to be a fabrication by AI, not Mr. Rosenbaum, quipping, “I also sound like I have a stick up my butt, according to ChatGPT.”
Another example comes from a chapter addressing the impact of social media and fabricated videos on teenagers. The book attributes two quotes to Lisa Feldman Barrett, a psychology professor at Northeastern University and author of “How Emotions Are Made: The Secret Life of the Brain.” The book quotes Ms. Barrett as writing, “Emotions aren’t just reactions to truth — they’re how we construct truth,” and “When young people say something ‘feels true,’ they’re describing a sophisticated process of meaning-making that integrates emotional and social signals.” However, Ms. Barrett informed The Times via email that these quotes “don’t appear in the book and they are also wrong.” She further clarified her academic stance, explaining, “I would never say ‘emotions aren’t just reactions to the truth’ — they are not reactions and ‘truth’ in science is a complicated concept that I tend to avoid,” adding, “Also, I would never say that ‘emotional and social signals’ are integrated — there are no emotional or social signals, per se. There are signals, and the brain creates their meaning as emotional or social.” This highlights not just misattribution but a misrepresentation of a scholar’s foundational academic principles.
Beyond outright fabrications, the book also contains genuine quotes that are improperly attributed or mixed with fake statements. One chapter cites Meredith Broussard, an NYU professor and author of “Artificial Unintelligence.” While the quote itself is authentic, it did not appear in her book; Broussard actually said it during a 2023 interview with “Marketplace Tech,” a daily radio show. Ms. Broussard commented, “It looks like this is either an A.I. hallucination or a misattributed quote.” Similarly, in a chapter exploring a “post-truth world” amplified by AI, the book quotes Lee McIntyre, a research fellow at Boston University, describing efforts to undermine truth as “‘a form of ideological supremacy,’ in which falsehood is used strategically — ‘not to misinform, but to displace truth as a societal value.'” McIntyre confirmed that while the first part of the quote was accurate, he had not said the second part verbatim, though he conceded that the general ideas were “concordant with my work.” He humorously noted, “It’s the ‘societal value’ part this looks wonky to me. One might say that about my work, without the quotation marks, and I think it would be OK. I just have never, to my knowledge, used that phrase.” These instances reveal a pattern of sloppy research, whether facilitated directly by AI or through a researcher’s reliance on AI.
In his comprehensive statement, Mr. Rosenbaum concluded by suggesting that if this episode “serves as a warning about the risks of A.I.-assisted research and verification, that is why I wrote the book.” He then pivoted, arguing that “These A.I. errors do not, in fact, diminish the larger questions that the book raises about truth, trust and A.I. and its impact on society, democracy and editorial.” While this statement attempts to salvage the book’s overarching message, the irony of a book about “The Future of Truth” grappling with issues of fabricated and misattributed information due to AI use is stark. This incident serves as a cautionary tale, not just for authors and publishers, but for anyone navigating the increasingly complex landscape of information where distinguishing human-generated content from AI-generated content is becoming a critical skill.